Vertical Structure of the Tropical Pipe in Austral Winter

Having established that the large-scale tropical leak events are observed at a range of levels, the vertical structure of the tropical pipe during austral winter is investigated in the follo- wing. Figure 3.14 shows vertical profiles of monthly mean REz10 for July of the years 2006,

2009, 2008, and 2010 (Figs. 3.14a-d, respectively), with monthly zonal mean N2O volume

mixing ratio contours (dotted green lines) and the average location of the subtropicalREz10

maximum for each hemisphere (red lines) superimposed.1 In both 2009 and 2010, the month of July lies in the middle of the leak period, though the leaks start earlier in 2010 (see also Fig. 3.12a). Above 700K,most of the leakage of air from the tropics has already occurred by July in 2010, which explains why no distinct high-REz10 structure is observed between 10°S

and 30°S that could represent the southern hemisphere tropical pipe edge (Fig. 3.14d). In July 2009 (Fig. 3.14b), a high-REz10 region is visible between 35°S and 40°S at rather higher

latitudes than usual for the tropical pipe edge in winter. This is similar to the situation found in 2005 (Fig. 3.7). The large amount of tropical air that has reached the mid-latitudes by this time has shifted the location of the N2O gradient poleward (see also Fig. 3.18 and Fig.

3.23b below). The flatness of the N2O mixing ratio contours in Figs. 3.14b and d between

the equator and the subtropical REz10 maximum illustrates how well-mixed the air is bet-

ween 800 and 1100 K.In contrast, the subtropicalN2O gradient of the southern hemisphere

tropical pipe edge lies much closer to the equator in July of 2006 and 2008 (between 15°S and 20°S, see Figs. 3.14a and c). At the centre of these gradients a vertically continuous band of high REz10 can be seen between 600 and 900 K in 2006 and up to 1000 K in 2008.

This confirms that the high latitudes at which the REz10 identifies the southern hemisphere

tropical pipe edge in some years (Fig. 3.11) are not an artefact of theREz10methodology but

are due to the large-scale leaks from the tropical pipe shifting the location of the maximum

N2O gradient poleward.

A high-REz10 structure probably associated with the N2O gradient at the polar vortex

edge is found near 60°S in all four years, but with some distinct differences between the years with and without large-scale tropical leaks. From 400 to 850 K, the polar high-REz10

region is centred close to 60°S in 2006 and 2008 (Fig. 3.14a and c). In 2009 and 2010, this is also the case below 550 K and the high-REz10 band is continuous up to at least 1000 K

with an equatorward tilt with altitude (Fig. 3.14b and d). Above 750 K in July 2010 the gradient between tropical and mid-latitudeN2O mixing ratios has been shifted so far toward

the polar vortex by tropical leakage that it cannot be distinguished from the gradient at the vortex edge anymore. Therefore the red line indicating the southern hemisphere subtropical

REz10 maximum in Fig. 3.14d cannot be considered a reliable estimate of the position of the

tropical pipe edge during this time. Its location is largely determined by the limitation to the region between 4.5°S and 40.5°S when searching for the REz10 maximum. In July 2009,

the southern tropical pipe edge and the polar high-REz10 region are distinct features. The

former is centred at 30°S between 650 and 1050 K, while the tilt of the latter means that it lies between 55°S and 48°S in this vertical range. These latitudes appear to be too low to be realistic, as was mentioned earlier.

1_{The years 2009 and 2010 were selected as an example for years when leaks are occurring as these were}

already examined in Sect. 3.2.2 and because 2005 and 2007 both have a data gap in July. 2006 and 2008 are the only available years during which no large-scale leakage is observed.

3.2. Analysis of the Tropical Pipe

Figure 3.14: Meridional profile of monthly average REz10 for July in (a) 2006 (b) 2009 (c)

2008, and (d) 2010. 2009 and 2010 are years during which large-scale leaks of air from the tropical pipe are observed, while 2006 and 2008 are not. Dotted green contours represent the monthly averageN2O volume mixing ratios inppbv. The red lines correspond to the average

location of the subtropical REz10maximum for each hemisphere and the error bars give the

standard deviation of the daily latitude of theREz10 maximum from the respective monthly

In 2006 and 2008, a high-REz10 region can be observed around 40°S above 900 K and

1000 K, respectively (Figs. 3.14a and c). At these levels the zonal mean N2O contours are

nearly horizontal from the equator to 30°S, particularly in 2006. This suggests that the high-

REz10 region near 40°S does represent the tropical pipe edge at higher altitudes and that

there is a sudden jump in the latitude of the tropical pipe edge compared to lower levels. The RE methodology correctly recognises this jump, as can be seen in the red line in the southern hemisphere in Figs. 3.14a and c. On the other hand, the polar high-REz10 region

only reaches up to approximately 850K in these years. This is probably related to a lowering of the signal-to-noise ratio as the volume mixing ratios decrease with altitude. In all four diagrams of Fig. 3.14, the regions of highREz10values tend to deteriorate above the 30 ppbv

contour. The single profile precision of the N2O observations at these altitudes is 13 ppbv

(Livesey et al., 2011). This implies that the signal-to-noise ratio above the 30ppbv contour is similar to that of theCH3Clobservations at 550K (see Sect. 3.1.2). Hence, the uncertainties

in the N2O observations are probably the reason why the RE cannot detect the gradients

at higher levels anymore. This also implies that the subtropical maxima of REz10 identified

by the red lines in Figs. 3.14a-d have to be considered unreliable estimates of the location of the tropical pipe edges above the 30 ppbv contour.

3.3

High-resolution Analysis of Tropical Leaks via Do-

main Filling

In order to investigate the large-scale leakages of tropical air into the mid-latitudes of the southern hemisphere (identified in Fig. 3.8) in more detail, a technique for artificially en- hancing the resolution of the available observations is used in the following. The technique is based on the reverse domain filling methodology introduced by Sutton et al. (1994). The number of data points for each day is increased by using a two-dimensional trajectory model (see Sect. 2.3.4) and reanalysis winds (MERRA, see Sect. 2.3.2) to advect tracer observations from previous and subsequent days to the central ‘current’ day. In particular, five days of observations on either side of the current one are advected to 12:00 (noon) UTC of that day. Observations for the central day that were not acquired at 12:00 UTC (±0 : 30, since the advection model works in hourly time steps) are also advected by the appropriate number of hours, thus creating a synoptic map of the tracer field with approximately eleven times as many data points than in the original observations. No more than five days on either side are used in order to minimise potential errors related to neglecting vertical motion.

Pierce et al. (1994) showed that the correlation between coincident adiabatically-advected and subsequent measurements of stratospheric tracers remains high for about ten days. The approach differs from the original reverse domain filling methodology (Sutton et al., 1994) in that the resulting synoptic tracer field is not a regularly sampled grid and because it also includes observations acquired after the central day. It will therefore be referred to simply as ‘domain filling’. Initially, the focus is on the leak occurring in 2009 as its evolution is representative for most other years and because it is one of the few years during which the EOS-MLS data has no gaps.